builtin terminfo: add custom 'query-os-name' #2217

krobelus commented 2025年11月12日 11:45:17 +01:00

Nice. We should probably make sure that terminals agree on the values for query-os_name.

I'd naively expect os.uname().sysname which is what uname prints,
so it should be a low common denominator (not specific to Python).
If we do that, we could call it query-uname which makes the semantics more obvious (and doesn't mix "-" and "_").

I'm not sure what are the downsides (compared to the kitty approach), maybe that the possible values are not known at compile time? Or that "darwin" is not as nice as "macos".
I'm also fine with platform.system().

Asked on the kitty tracker -> https://github.com/kovidgoyal/kitty/issues/9217

Nice. We should probably make sure that terminals agree on the values for `query-os_name`. I'd naively expect `os.uname().sysname` which is what `uname` prints, so it should be a low common denominator (not specific to Python). If we do that, we could call it `query-uname` which makes the semantics more obvious (and doesn't mix "-" and "_"). I'm not sure what are the downsides (compared to the kitty approach), maybe that the possible values are not known at compile time? Or that "darwin" is not as nice as "macos". I'm also fine with `platform.system()`. Asked on the kitty tracker -> https://github.com/kovidgoyal/kitty/issues/9217

dnkl commented 2025年11月12日 11:48:18 +01:00

I would also prefer uname. This was really just a quick hack to show how it can be done.

I would also prefer uname. This was really just a quick hack to show how it can be done.

dnkl commented 2025年11月12日 11:48:45 +01:00

+1 for calling it query-uname as well :)

+1 for calling it `query-uname` as well :)

dnkl commented 2025年11月12日 13:21:18 +01:00

Even if we keep the original name, query-os_name (query-os-name? query-os?), we could still standardize on using uname as source. That would make it very easy for all Unix terminals. Windows terminals would have to decide together, which name to use.

Even if we keep the original name, `query-os_name` (`query-os-name`? `query-os`?), we could still standardize on using `uname` as source. That would make it very easy for all Unix terminals. Windows terminals would have to decide together, which name to use.

dnkl commented 2025年11月12日 13:23:39 +01:00

I've updated the MR to use uname.sysname instead.

I've updated the MR to use `uname.sysname` instead.

krobelus commented 2025年11月12日 16:36:15 +01:00

To be explicit about the data source, we could add a uname= prefix (query-os-name=uname=$(uname)).
I guess conflicts would be quite unlikely but this seems a bit cleaner.
I'm okay either way I've cc'ed some folks in https://github.com/wezterm/wezterm/issues/7370 in case others want share thoughts.

I wouldn't call .lower() because that makes it needlessly different from uname(1).

Also some nits

• I'd change query-os_name to query-os-name though I'm fine with either
• (I generally don't like abbreviations in globally visible name so I wonder if query-system-name would be better.) Hard to tell, so I'd leave query-os.

To be explicit about the data source, we could add a `uname=` prefix (`query-os-name=uname=$(uname)`). I guess conflicts would be quite unlikely but this seems a bit cleaner. I'm okay either way I've cc'ed some folks in https://github.com/wezterm/wezterm/issues/7370 in case others want share thoughts. I wouldn't call `.lower()` because that makes it needlessly different from `uname(1)`. Also some nits - I'd change `query-os_name` to `query-os-name` though I'm fine with either - (I generally don't like abbreviations in globally visible name so I wonder if `query-system-name` would be better.) Hard to tell, so I'd leave `query-os`.

dnkl commented 2025年11月19日 08:31:23 +01:00

Updated to:

• query-os-name
• Use uname -s as is (no lower-casing)
• Prefix the result with uname=

$ ./utils/xtgettcap query-os-name
reply: (56 chars): <ESC>P1+r71756572792d6f732d6e616d65=756E616D653D4C696E7578<ESC>\
 query-os-name=uname=Linux

Updated to: * `query-os-name` * Use `uname -s` as is (no lower-casing) * Prefix the result with `uname=` ``` $ ./utils/xtgettcap query-os-name reply: (56 chars): <ESC>P1+r71756572792d6f732d6e616d65=756E616D653D4C696E7578<ESC>\ query-os-name=uname=Linux ```

krobelus commented 2025年11月19日 14:17:59 +01:00

Thanks! Tested successfully with this upcoming change.

Usage would be a bit simpler without the uname= prefix but I think I still prefer its explicitness.

So this LGTM; maybe mention it in docs somewhere like this:

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 6dd0719fa3..b4aec92c16 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -74,8 +74,8 @@
 `foot` and `footclient`), allowing you to set a custom toplevel
 tag. The compositor must implement the new `xdg-toplevel-tag-v1`
 Wayland protocol ([#2212][2212]).
-* `XTGETTCAP`: added `query-os_name`, returning the OS foot is
- compiled for (e.g. _'linux'_) ([#2209][2209]).
+* `XTGETTCAP`: added `query-os-name`, returning the OS foot is
+ compiled for (e.g. _'Linux'_) ([#2209][2209]).
 
 [2212]: https://codeberg.org/dnkl/foot/issues/2212
 [2202]: https://codeberg.org/dnkl/foot/issues/2202
diff --git a/README.md b/README.md
index e8f3c8cd5a..632e6147f3 100644
--- a/README.md
+++ b/README.md
@@ -641,6 +641,9 @@
 capability name, foot's reply is identical to what `tigetstr()` would
 have returned.
 
+In addition to queries for terminfo entries, the `query-os-name`
+query will be answered with a response of the form `uname=$(uname)`,
+where `$(uname)` is the name of the OS foot was compiled for.
 
 # Credits
 

Thanks! Tested successfully with [this upcoming change](https://github.com/fish-shell/fish-shell/commit/14414ceec18fb836e61132dde1a9b4404d2e3fc9). [Usage](https://github.com/fish-shell/fish-shell/commit/14414ceec18fb836e61132dde1a9b4404d2e3fc9#diff-d8281319c42f0263bf9edf56669da239780d8e39ffee494bb4b7e06ba2303746R2) would be a bit simpler without the `uname=` prefix but I think I still prefer its explicitness. So this LGTM; maybe mention it in docs somewhere like this: ```diff diff --git a/CHANGELOG.md b/CHANGELOG.md index 6dd0719fa3..b4aec92c16 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -74,8 +74,8 @@ `foot` and `footclient`), allowing you to set a custom toplevel tag. The compositor must implement the new `xdg-toplevel-tag-v1` Wayland protocol ([#2212][2212]). -* `XTGETTCAP`: added `query-os_name`, returning the OS foot is - compiled for (e.g. _'linux'_) ([#2209][2209]). +* `XTGETTCAP`: added `query-os-name`, returning the OS foot is + compiled for (e.g. _'Linux'_) ([#2209][2209]). [2212]: https://codeberg.org/dnkl/foot/issues/2212 [2202]: https://codeberg.org/dnkl/foot/issues/2202 diff --git a/README.md b/README.md index e8f3c8cd5a..632e6147f3 100644 --- a/README.md +++ b/README.md @@ -641,6 +641,9 @@ capability name, foot's reply is identical to what `tigetstr()` would have returned. +In addition to queries for terminfo entries, the `query-os-name` +query will be answered with a response of the form `uname=$(uname)`, +where `$(uname)` is the name of the OS foot was compiled for. # Credits ```

dnkl commented 2025年11月20日 13:59:55 +01:00

Applied, thanks!

Applied, thanks!

krobelus commented 2025年11月23日 19:43:01 +01:00

we got a vote for removing the uname= prefix (https://github.com/wezterm/wezterm/issues/7370#issuecomment-3567974927),
so I think we should do that (it's less surprising).

we got a vote for removing the `uname=` prefix (https://github.com/wezterm/wezterm/issues/7370#issuecomment-3567974927), so I think we should do that (it's less surprising).

dnkl commented 2025年12月10日 13:45:36 +01:00

I'm fine with either. I've dropped the uname=-prefix.

I'm fine with either. I've dropped the `uname=`-prefix.